[R] adding vertical segments to an xyplot in lattice

[Peter Ehlers]

On 2011-03-22 12:12, Christopher W Ryan wrote:
> I have a dataframe that looks like this:
>
>   >  str(chr)
> 'data.frame':   84 obs. of  7 variables:
>   $ county: Factor w/ 3 levels "Broome","Nassau",..: 3 3 3 3 3 3 3 3 3 3 ...
>   $ item  : Factor w/ 28 levels "Access to healthy foods",..: 21 19 20
> 18 16 3 2 6 17 8 ...
>   $ value : num  8644 15 3.5 3.9 7.7 ...
>   $ low   : num  7897 9 2.5 2.6 7 ...
>   $ high  : num  9390 22 4.5 5.2 8.4 37 30 23 24 101 ...
>   $ target: num  5034 11 2.7 2.6 6.1 ...
>   $ nys   : num  6099 16 3.5 3.3 8 ...
>
>> head(chr)
>      county                      item  value    low   high target    nys
> 1 Sullivan           Premature death 8644.0 7897.0 9390.0 5034.0 6099.0
> 2 Sullivan       Poor or fair health   15.0    9.0   22.0   11.0   16.0
> 3 Sullivan Poor physical health days    3.5    2.5    4.5    2.7    3.5
> 4 Sullivan   Poor mental health days    3.9    2.6    5.2    2.6    3.3
> 5 Sullivan           Low birthweight    7.7    7.0    8.4    6.1    8.0
> 6 Sullivan             Adult smoking   29.0   22.0   37.0   15.0   20.0
>
> I'd like to graph high and low for "Premature death" for each of the
> three counties, with 3 vertical line segments, one connecting those
> two points for each county.  I can get the two points for each county:
>
>> xyplot(low+high ~ county, data=subset(chr, item=="Premature death"))
>
> but I have not yet been able to figure out how to draw the 3 vertical
> line segments. Been struggling to understand panel functions, but no
> success so far. I'd be grateful for any advice.

For lattice, I usually prefer the long version of a dataset.
Try this:

dd <- data.frame(county = letters[1:3],
                  lo = c(5,2,3),
                  hi = c(9,5,10))

## convert to 'long' format (you can use the reshape() function in stats 
or the reshape package:

require(reshape)
dd.long <- melt(dd, id = "county")
dd.long

require(lattice)
xyplot(value ~ county, data = dd.long, groups = county,
   pch = 19, type = 'b', cex = 2, lwd = 5, col = 2:4)

Peter Ehlers

>
> Thanks.
>
> --Chris Ryan
> SUNY Upstate Medical University
> Clinical Campus at Binghamton